Can We Resurrect the CAPM in Japan? Evaluating Conditional Asset Pricing Models by Incorporating Time-varying Price of Risk

This paper evaluates conditional asset pricing models for the Japanese stock market by examining time-varying risk pricing. Using a multivariate generalized autoregressive conditional heteroskedasticity (GARCH) model, we tested the conditional versions of the Sharpe (1964) Lintner (1965) Mossin (1966) capital asset pricing model (CAPM), the consumption CAPM (CCAPM), and the CAPM with a constant term. The empirical results demonstrate that the price of risk in the conditional CAPM is generally positive and significant. Moreover, our formal panel data tests reveal that the conditional CAPM is never rejected in the case of the 25 book-equity-to-market equity (BE/ME)-ranked portfolios. Furthermore, in the conditional version of the CAPM with a constant term, positive alphas are generally seen in Japan; however, our statistical test of the hypothesis that the average value of the alphas of the conditional CAPM equals zero is never rejected for the 25 BE/ME-ranked portfolios in Japan. This evidence demonstrates that the CAPM can be adequately represented by using the multivariate GARCH model to explain the value premia in Japan.